Unmanned Systems Technology 027 l Hummingbird XRP l Gimbals l UAVs insight l AUVSI report part 2 l O’Neill Power Systems NorEaster l Kratos Defense ATMA l Performance Monitoring l Kongsberg Maritime Sounder

80 In operation | Kratos Defense ATMA fusion algorithms in-house, combining the driver software from different ATMA component manufacturers with ROS (robotic operating system). “ROS has a lot of open-source modular components that we’ve integrated into the software,” Factor says. “That keeps things standardised and eliminates anyone’s special coding techniques, which is important for evaluation and keeping the system safe.” Perhaps most important, the ATMA will not react to fast incoming vehicles from the rear. The impact protection payload being pulled by the follower truck must remain in place to protect the work zone. If a crash should occur, it will be detected either by an impact sensor in the crash attenuator or by the SCU detecting a jolt in the follower vehicle’s GNSS coordinates that is characteristic of a collision. In response, the follower immediately activates its brakes to avoid rolling forwards into the workspace. As Factor explains, “In many real-world cases of highway maintenance work accidents, the human driver has had the instinct to get out of the way when they see an incoming large vehicle in the rearview mirror, and the impact will clip the side and send the vehicle flying down it. Our automation system eliminates that altogether – you don’t have to worry about that, because automated systems don’t get scared.” Follow-up analysis and future research When the highway work team has finished a job, the ATMA operator can switch off the Follow function and move into the follower vehicle. They then switch it back into manual mode before driving it (typically with the lead vehicle) to wherever it is needed next. “Our SCU’s SD card is capable of storing up to 24 hours of data, which can then be analysed afterwards to improve future operations,” says Factor. That means the operating team can judge the overall safety of the ATMA system, for example by seeing how closely the vehicles were aligned during operation. With post-processed kinematic GPS, the exact paths taken by the leader and follower trucks can be plotted with centimetre-level accuracy, and analysed for inconsistencies as small as a few centimetres. While that could be time- consuming and potentially unnecessary for most operations, it may be that an accident or a problem with the system occurred during a day’s work, so an end-user might run such an analysis to assess why it happened. In some locations, Kratos is also supporting a research-oriented pilot programme. As Factor explains, “That’s necessary for legislators and authoritative bodies to approve this system, to let us use it on state roads. They want to be able to see how accurate it really is, or if there were any kind of obstacles disrupting the operation, and similar sorts of variables. They’ll use that data for research purposes, but in normal operation, users have no major need for the data.” Conclusion As more approvals and customer feedback are gained, the ATMA system is being continually redesigned and improved to suit different trucks and transport regulations, enabling more and more lives to be saved on highways around the world. August/September 2019 | Unmanned Systems Technology Vehicle chassis: Royal Truck & Equipment Main computer: in-house Operating system: Windows 10 Servo actuators: Moog GNSS-INS: Trimble GNSS antennas: Trimble Cameras: SV3C Radars: Delphi Lidars: LeddarTech Main and redundant data links: Microhard Independent emergency stop data link: FreeWave Technologies Operator control system: Panasonic Some key suppliers Ultrasonic sensors provide close-range collision warnings, as is standard for automotive sensors